![mindsync on Twitter: "The new challenge is to create and train massive neural networks and then optimize those networks to run efficiently on a DSA, be it a CPU, GPU, TPU, ASIC, mindsync on Twitter: "The new challenge is to create and train massive neural networks and then optimize those networks to run efficiently on a DSA, be it a CPU, GPU, TPU, ASIC,](https://pbs.twimg.com/media/EGWFZiIWkAAywhb.jpg)
mindsync on Twitter: "The new challenge is to create and train massive neural networks and then optimize those networks to run efficiently on a DSA, be it a CPU, GPU, TPU, ASIC,
A Survey Paper Comparing modern CPU, GPU & TPU Hardware in Relation to Neural Network Training and Inference. | by T Robert Allred | Medium
![urban liebel on Twitter: "CPU, GPU, TPU and now HIVE, makes total sense ( and no it wasn't me , I swear ;-) https://t.co/PM1VmEm8TT" / Twitter urban liebel on Twitter: "CPU, GPU, TPU and now HIVE, makes total sense ( and no it wasn't me , I swear ;-) https://t.co/PM1VmEm8TT" / Twitter](https://pbs.twimg.com/media/DCI_H4-XkAA2DIc.jpg)
urban liebel on Twitter: "CPU, GPU, TPU and now HIVE, makes total sense ( and no it wasn't me , I swear ;-) https://t.co/PM1VmEm8TT" / Twitter
![CPU vs GPU vs TPU and Feedback Loops in real world situations | by Inara Koppert-Anisimova | unpackAI | Medium CPU vs GPU vs TPU and Feedback Loops in real world situations | by Inara Koppert-Anisimova | unpackAI | Medium](https://miro.medium.com/max/500/1*P7Jyst9ASw6n-0JSkx9SAA.jpeg)
CPU vs GPU vs TPU and Feedback Loops in real world situations | by Inara Koppert-Anisimova | unpackAI | Medium
![Google says its custom machine learning chips are often 15-30x faster than GPUs and CPUs | TechCrunch Google says its custom machine learning chips are often 15-30x faster than GPUs and CPUs | TechCrunch](https://beta.techcrunch.com/wp-content/uploads/2017/04/2017-04-05_1014.png)